Tracking of Fish School in Aquaculture with Weighted Clustering Technique Using Kalman Filter

Abstract

Behavior of fish school can provide reliable inference about the health of fishes and their aquaculture environment. Monitoring the fish activity demands an intelligent system, which has been addressed in this work. The proposed work uses an overhead vision camera to track the activity of the fish school in an aquaculture tank. Problems associated with the occlusion of fishes are addressed using the weighted K-means clustering technique. It updates weights to the fish occupied region based on their area and is clustered using these weights, which provides an accurate estimate of the center of the fish school. Temporal variations of these fish schools are tracked using the Kalman filter-based multi-target tracking approach. It manages the tracks based on the associated fish schools observed using the vision camera. Experimental results illustrate the reliability of the proposed technique in monitoring fish school activity in an indoor aquaculture environment.